<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Google DeepMind on Bitsy Wiki</title>
    <link>https://wiki.bitsy.services/wiki/ai/models/google/</link>
    <description>Recent content in Google DeepMind on Bitsy Wiki</description>
    <generator>Hugo</generator>
    <language>en</language>
    <atom:link href="https://wiki.bitsy.services/wiki/ai/models/google/index.xml" rel="self" type="application/rss+xml" />
    <item>
      <title>Gemini 3 Pro</title>
      <link>https://wiki.bitsy.services/wiki/ai/models/google/gemini-3-pro/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      <guid>https://wiki.bitsy.services/wiki/ai/models/google/gemini-3-pro/</guid>
      <description>&lt;p&gt;Gemini 3 Pro is the base model of &lt;a href=&#34;https://wiki.bitsy.services/wiki/ai/models/google&#34;&gt;Google DeepMind&lt;/a&gt;&amp;rsquo;s third Gemini generation, released November 2025. It earns a page not because it is the newest model Google offers — it is not — but because &lt;strong&gt;every Gemini model Google shipped in 2026 descends from it&lt;/strong&gt;, and its model card is the only document in the family that describes the architecture at all.&lt;/p&gt;&#xA;&lt;p&gt;Its own card states that it is not itself derived from anything: &amp;ldquo;Gemini 3 Pro is not a modification or a fine-tune of a prior model.&amp;rdquo;&lt;/p&gt;</description>
    </item>
    <item>
      <title>Gemma</title>
      <link>https://wiki.bitsy.services/wiki/ai/models/google/gemma/</link>
      <pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
      <guid>https://wiki.bitsy.services/wiki/ai/models/google/gemma/</guid>
      <description>&lt;p&gt;Gemma is &lt;a href=&#34;https://wiki.bitsy.services/wiki/ai/models/google&#34;&gt;Google DeepMind&lt;/a&gt;&amp;rsquo;s open-weight model line: the weights are published as files, and anyone can download them and run them on their own machines. It exists alongside Gemini, which is closed-weight and reachable only through Google&amp;rsquo;s interfaces. The two share research but not checkpoints — Google describes Gemma as &amp;ldquo;built from the same technology that powers our Gemini models,&amp;rdquo; not as a smaller Gemini.&lt;/p&gt;&#xA;&lt;p&gt;The line matters for a reason that has nothing to do with capability rankings. An open-weight model can process data that must not leave a building, can be fine-tuned by a third party, cannot be withdrawn by its publisher, and can be inspected by anyone. Gemma is Google&amp;rsquo;s answer to all four, and since April 2026 its licence stops qualifying any of them.&lt;/p&gt;</description>
    </item>
  </channel>
</rss>
